Caret5 Spherical Registration Core 6 Landmark Set

2010-03-03, Donna Dierker, Van Essen Lab

Note: Most captures are much larger than they appear on this page. Mozilla has a View Image option when you right-click an image; try right-clicking an image to see if your browser has a similar option for viewing a specific capture in full size.

List of Core 6 Landmarks

The core 6 landmarks include:

  1. central sulcus (CeS) (quick ref: 18mm off dorsal; 19mm off ventral)
  2. sylvian fissure (SF) (quick ref: 9mm posterior to the intersection with the circular sulcus on FLAT)
  3. anterior half of the superior temporal gyrus (aSTG)
  4. calcarine sulcus (CaS) (quick ref: 24mm from occipital pole)
  5. medial wall dorsal segment (quick ref: 19mm gap at frontal junction)
  6. medial wall ventral segment (quick ref: 16mm gap at calcarine junction)

Links to Datasets

McLaren & Burton Guidelines

In July 2006, Donald McLaren and Harold Burton developed a more detailed version of the core 6 landmarks that attempts to tighten up potential inter-rater variance. Neuroanatomists likely will prefer the McLaren/Burton guidelines, while others may need the help of an atlas in order to follow the anatomical landmarks. A rigorous comparison of the original and McLaren/Burton methods remains to be done, but three raters drew borders on two hemispheres before and after using these guidelines. Inter-rater dispersion along the medial wall ventral segment, from the amygdala to the olfactory sulcus, was substantially reduced after using these guidelines, but medial wall dorsal dispersion was worse. We recommend using McLaren/Burton guidelines for the medial wall ventral segment; for the medial wall dorsal, just trace along the dark line corresponding to the callossal sulcus, when it is clearly defined. When it is not clearly defined, and the problem is not poor peak selection, then use the McLaren/Burton method. You can find the McLaren/Burton guidelines here:

McLaren/Burton Core 6 Landmark Guidelines

Other Useful Links

This document supersedes the documents below, wherever they conflict:

Erin Reid's Caret Instructions for Windows (Word document); aka "protocol" or "cheat sheet"
Caret 5.2 Tutorial - Introduction, Installation, Visualization, and Registration (PDF)
Caret 5.2 Tutorial - Segmentation and Flattening (PDF)
Dataset for Caret Advanced Tutorial at Wash U June 2006 (zip archive contains Word doc in DOCUMENT folder)

General Guidelines

The extent of each landmark is constrained to regions where it is reasonable to expect good correspondence across nearly all subjects. Where distances are used, the reference surface is specified. The inflated surface is the reference surface for all landmarks except the sylvian, for which the flat surface is used; make sure you use the reference surface when checking ID distances. By "inflated," we mean the SureFit-generated (inflate_surface with smooth_iters1=60; smooth_factor1=1.0; num_cycles=6; inflate_factor=0.6), moderately inflated surface -- not VERY_INFLATED or Freesurfer (highly) inflated. Distances are +/- 10%. If your individual's brain is sufficiently bigger or smaller than 711-2C (wustl.edu atlas space used for PALS_B12, Colin -- closer to MNI305 than Talairach, but not quite as big as MNI305), then you might need to use your judgment. If using these distances to determine starting and end points results in borders that look disproportionate to Colin's (e.g., posterior-most point of calcarine border is closer to medial wall than occipital pole for a very small brain), then adjust the distances to better match proportions. At this point, we're not recommending you compute the ratio of your subject's surface area or volume to colin's and reduce your distances based on the resulting geometric distance ratio; rather, "eyeball it" and make sure there's a logical explanation for large deviations from these reference distances (e.g., very big or small brain). Ultimately, the most important rule is to draw landmarks consistently from one subject to the next. Minor deviations from these guidelines are relatively unimportant if they are applied consistently. Of course, the worst deviations are those applied to one group of subjects, but not another, if the goal is to compare results across those groups. Consistency is essential.

Since the template cuts often are redrawn during flattening, many of the inflated and spherical views shown were captured with the CUT topo file assigned to show the starting/termination points of these cuts. During normal flattening and registration, the CLOSED topo file should be assigned to all 3-D surfaces; the CUT topo file should be used only with the flat map.

Extent Guidelines using Colin Reference

Note: While we use Colin as an example target when drawing landmark borders, the 'mental target' is to match the average values represented by the PALS_B12 Case 1-12 averages. The extent guidelines herein reflect those averages, which may differ from those seen in the colin reference datasets by a few mm (for details, see the case-by-case extent reference measurements, averages, and standard deviations for both hemispheres in PALS_B12_Core6_LANDMARK-measurements_11May05.xls).

CeS extent: Rotating as needed to approximate AC-PC alignment, the central sulcus landmark border begins about 19mm above the margin of the sylvian fissure (distance reference: inflated surface). It terminates about 18mm from the medial wall dorso-medially (distance reference: inflated surface). (Note: Before 5/11/2005, the guideline was 15mm from each end.)

LeftRight

SF extent: The sylvian fissure landmark begins along the posterior segment of the Sylvian fissure, starting about 9mm posterior to the intersection with the circular sulcus (whose inferior and superior branches merge at this locus). It then continues along the superior branch of the circular sulcus. In the screen captures below (Flat map and Inflated map), green ID nodes mark the inferior branch of the circular sulcus. The border should be drawn and the 9 mm distance based on the flat surface. However, you should open a second window and load the inflated surface to ensure that you have correctly identified the inferior branch of the circular sulcus; using the flat map alone can be unreliable. On the inflated map this intersection is just before the beginning of the posterior branch of the sylvian fissure and appears slightly posterior to the gyral inflation that is just posterior to the postcentral sulcus. Anteriorly, the landmark extends almost to the anterior and ventral limit of the superior branch of the circular sulcus, 10 mm dorsal to the ventral margin of the frontal lobe. (Before 10/5/2005, we identified this intersection as the "SF-SF2 intersection, where we named SF the sylvian fissure's primary fundus and SF2 the main secondary fundus. From 12/21/2004 until 5/11/2005, the guideline was 12mm posterior to the SF-SF2 intersection.) Before 12/21/2004, Donna Hanlon aimed for the SF-SF2 intersection, requiring rework of borders on CSM and Amaral studies.

FlatAntero-Lateral Inflated

aSTG (aka SF_STSant aka "the pink one") extent: Rotating as needed to approximate AC-PC alignment, drop a plumb bob down from the the central sulcus border's ventral tip until it hits the STG (i.e., directly below the CeS ventral tip -- not projecting along the path of the CeS): This is your starting point. Terminate it at the anterior limit of the temporal pole.

AnteriorLateralSphere

CaS extent: In Caret, this border is drawn as a cut during flattening. Erin Reid draws the calcarine from the medial wall outward to the occipital pole, while Donna Hanlon draws it from the occipital pole to the medial wall. You can draw it either way during flattening; Caret will reverse its polarity if you draw it from the medial wall to the occipital pole. For registration purposes, it commences 24mm from the occipital pole on the inflated surface and intersects the medial wall. In some subjects, the calcarine sulcus forks/splits; in such cases, draw along the ventral tine of any split/fork in the sulcus. An exception is if the ventral tine takes a hard ventral turn; you don't want a sharp turn in the calcarine border. In such cases, instead draw a border that extends from the occipital pole to the medial wall along as much of the CaS as possible. If your subject's calcarine sulcus is interrupted by a gyrus ("speed bump"), then draw the border across the gyrus. When drawing the border during flattening, extend the border to the occipital pole; however, during registration, nibble off border points at the posterior end until there is a 24mm gap on the inflated surface between the occipital pole and the posterior end of the CaS landmark border. Click first on posterior view; ID occipital pole; switch to medial view; ID point about 24mm anterior; distance reference: inflated surface. Before 5/11/2005, the guideline was 27mm gap (28 mm on the left side, 20 mm on the right side).

PosteriorMedial

Medial wall dorsal and ventral segments: The dorsal medial wall landmark runs along the corpus callosum, from a starting point just posterior to the olfactory sulcus to a termination just anterior to the calcarine sulcus. The ventral medial wall landmark runs from the same termination points, but in the opposite direction and along the medial margin of the hippocampal sulcus (15 mm medial to the parahippocampal gyrus at its maximum), and across the margins of cortex with the basal forebrain. Again, these borders normally are drawn during flattening as a single, closed border with no gaps. During flattening, it doesn't matter where the user starts and stops drawing, so long as the border is closed. Caret splits this border into dorsal and ventral segments, based on where the CaS (dorsal ends, ventral starts) and frontal cut (dorsal begins, ventral ends). As long as the frontal cut intersects the medial wall posterior to the olfactory sulcus (see green ID nodes below), it is not necessary to redraw the template cut; however, if it intersects dorsal/medial to the gyrus rectus or ventral/lateral to the medial orbital gyrus, then redraw the frontal cut so that it intersects the medial wall posterior to the olfactory sulcus (see below). During registration, nibble off border points near the two junctions so that the gaps look similar to those in the colin surfaces (19mm gap on the inflated surface at the frontal junction; 16mm at the calcarine junction). Doing so not only reduces the likelihood of crossovers along the medial wall ventral segment, but also ensures your borders are consistent with the many cases Erin has registered for various projects.

LeftRight

Borders for Three Amaral Datasets

When drawing registration borders for your individual, you should always view the landmark COLIN.L_REF-for-LANDMARKS or COLIN.R_REF-for-LANDMARKS dataset in a separate caret session, so you can reference its landmarks. Even then, it can be easier said than done to draw the corresponding borders on your subject. Below are examples of flattening and registration borders for three left hemispheres generously provided by David Amaral's Lab at UCDavis.

1939.L

This subject was rotated slightly chin-up from AC-PC line, so getting the aSTG right required rotating somewhat about Z. This rotation also caused trouble for the template cuts. The frontal cut intersected the medial wall just ventral to the medial orbital gyrus, and the sylvian cut intersected the aSTG, so I redrew all cuts.

Template Flattening BordersRedrawn Flattening Borders

The calcarine border is the same in both medial views below; it begins where it should for registration purposes (i.e., you could have drawn flattening cut further posterior, to occipital pole, and then nibbled off border points when drawing registration borders, so that it matches the extent shown below). The medial wall border is shown on the left with no gaps (i.e., how it looks when you draw it for flattening) and on the right with the gaps at the frontal and calcarine junctions (i.e., for registration). Interestingly, in this case no border points were deleted; the gaps were about the right distance "as is" (actually a bit too big at the calcarine junction, but not worth moving border points).

Flattening (no gap)Registration (with junction gaps)

Here are views of the other registration borders:

Flat view of Ces, SF, and aSTGLateral view of Ces, SF, and aSTGDorsal view of Ces

2191.L

Here are this subject's calcarine and medial wall borders:

Medial Wall at Flattening (no gaps)Medial Wall at Registration (with gaps)Medial-Ventral View of Gaps

Here are views of the other registration borders:

Flat view of Ces, SF, and aSTGLateral view of Ces, SF, and aSTGDorsal view of Ces

7382.L

This subject's template cuts didn't really need to be redrawn, but they are redrawn here for illustration purposes:

Compressed Medial Wall ViewView of Cut Origination
Ventral View of CutsView of Cut Termination

Here are this subject's calcarine and medial wall borders:

Medial Wall at Flattening (no gaps)Medial Wall at Registration (with gaps)Medial-Ventral View of Gaps

Here are views of the other registration borders:

Flat view of Ces, SF, and aSTGLateral view of Ces, SF, and aSTGDorsal view of Ces